2009/7/7 Duncan Coutts :

On Mon, 2009-07-06 at 18:28 -0700, John Meacham wrote:

...

Well, without a replacement, it seems odd to remove it. Also, Haskell currently doesn't _have_ a record syntax (I think it was always a misnomer to call it that) it has 'labeled fields'. None of the proposed record syntaxes fit the same niche as labeled fields so I don't see them going away even if a record syntax is added to haskell in the future.

The people proposing this can correct me if I'm wrong but my understanding of their motivation is not to remove record syntax or immediately to replace it, but to make it easier to experiment with replacements by making the existing labelled fields syntax a modular part of the language that can be turned on or off (like the FFI).

I'm not sure that I agree that it's the best approach but it is one idea to try and break the current impasse. It seems currently we cannot experiment with new record systems because they inevitably clash with the current labelled fields and thus nothing changes.

I think it is a powerful approach to try and break the current impasse for the following reasons: 1. Once implemented, Hackage and Cabal will soon give us accurate data on what publicly available Haskell code does and does not depend on NamedFields/TraditionalRecordSyntax/WhateverWeEndUpCallingIt 2. Once deprecated, people will be encouraged to not depend on the traditional record syntax where the cost of avoiding it is small (I'm thinking of situations like the mtl-accessors / run functions where the traditional syntax is saving something like one function definition). 3. Champions of alternative record syntaxes will know what on Hackage they can use out-of-the-box and what things they'd want to consider re-writing as examples of how their approach is superior. Does anyone have a concrete dea of what it would take to carve out the existing syntax as an addendum? Thanks, - Ravi

On Tue, Jul 07, 2009 at 10:28:11AM -0400, Ravi Nanavati wrote:

2. Once deprecated, people will be encouraged to not depend on the traditional record syntax where the cost of avoiding it is small (I'm thinking of situations like the mtl-accessors / run functions where the traditional syntax is saving something like one function definition). 3. Champions of alternative record syntaxes will know what on Hackage they can use out-of-the-box and what things they'd want to consider re-writing as examples of how their approach is superior.

Well, that's the problem. I don't want to ever deprecate the named field mechanism as it is completely independent of a record system. You can think of the possibilites as follows. named positional declared LabeledFields DataDeclarations anonymous ???? Tuples So, a record syntax fills the hole of anonymous, named field types. However, we shouldn't do it at the expense of opening another hole with declared, named field types any more than we should dispose of standard data declations in favor of making everything newtypes of tuples. Also, this supports the idea of using parethensis rather than braces for a record mechanism. this would be consistent with tuples in that parens = anonymous. Also, if you just replace braces with parens in the scoped records paper, you get no conflicts with existing syntax. (including named fields) :) John -- John Meacham - ⑆repetae.net⑆john⑈ - http://notanumber.net/

Malcolm Wallace wrote:

On 7 Jul 2009, at 02:28, John Meacham wrote:

...

Haskell currently doesn't _have_ a record syntax (I think it was always a misnomer to call it that) it has 'labeled fields'. ...

and a reworking of the standard to not refer to the current system as a 'record syntax' but rather a 'labeled fields' syntax.

I strongly agree with the latter. In fact, I was under the impression that the Report already avoided the term "record syntax" completely, but checking just now showed 6 distinct occurrences.

ah hah. Existing extension names: NamedFieldPuns (was erroneously "RecordPuns" in GHC for a release) RecordWildCards, DisambiguateRecordFields this "extension" could be named NamedFields. (then giving the lie to the above new names which maybe ought to be more like FieldWildCards and DisambiguateNamedFields(DisambiguateFieldNames?)) Also there is "ExtensibleRecords" which I guess refers to Hugs' TRex? -Isaac